login hangs on HP 9000/800 ops HP-UX B.11.00

[Qingfeng Zhong]

This is long standing issue happened most of the HP boxes.

It seems a progressive issue. Reboot of the boxes will solve the problem and keep the boxes going for a while.

The basic symptom is that login hangs, if one have HISTFILE pointed to HOME dir (which is a NFS mount). Repointed HISTFILE to /tmp (which is a local disk) would solve the problem. However, other problems may occur afterwards.

Further investigation shows that, when hangs, the client side is waiting for a signal back from its child. At server side, telnetd daemon dies and leave ksh shell be a orphan.

Can anybody tell me why telnetd die in these cases? How can we fix it?

[Qingfeng Zhong]

I just want to add:

When login hangs, the involved NFS mount is working.

The hanging goes forever.

I needs to find a root cause of this.

[Cheryl Griffin]

Login hanging when the home directory is an NFS mount is a very common issue. It can indicate a problem with the network or network services. The home directory on the NFS mount may not be responding as quickly as it should.

A properly patched system will lessen these hangs.

PHKL_23409 NFS, Large Data Space, kernel memory leak
PHNE_27902 Cumulative STREAMS Patch
PHNE_29473 cumulative ARPA Transport patch
PHNE_26096 telnet kernel and telnetd(1M) patch
PHCO_27721 login(1) cumulative patch

Do the hangs still occur if the system is properly patched?

[Qingfeng Zhong]

Thanks, Cheryl,

I will ask our sys admin to patch it.

There is still one thing puzzles me. If it is the network issue, the behavior should be sporadic. By which, I mean when the network is good it should works, it only goes bad when the network is not available. When the network comes back later, it should work again. However, this is not what we observed. In our case, it is more like a degrading of computer. By which, I mean, it works after reboot of the computer. It will work for a while until certain point. After that, it will not work unless we reboot the box.

This reason I am asking this here is that we observe lots of weired behavior after we have the login issue. I am wondering if they are related degrading of computer and could be fixed by the patches, too, or not.

Thanks again.

[Elmar P. Kolkman]

One more thing. Just ran into this myself.

HP-UX 11i allows to mount NFS volumes even though /etc/rc.config.d/nfsconf is not configured for an NFS client, and thus the processes like rpc.statd and rpc.lockd are not running...

By configuring nfsconf to be at least an NFS client, and starting /sbin/init.d/nfs.client, the blocking was solved in my case. Perhaps in yours too ?